The present invention generally relates to semiconductor materials and isoelectronic surfactant-induced sublattice disordering in optoelectronic devices and, more particularly, to a method of surfactant-induced sublattice disordering for solar conversion and other photovoltaic devices.
The interest in photovoltaic (PV) cells in both terrestrial and non-terrestrial applications continues as concerns over pollution and limited resources increase. Irrespective of the application, and as with any energy generation system, efforts have been ongoing to increase the output and/or increase the efficiency of PV cells. To increase the electrical power output of such cells, multiple subcells or layers having different energy band gaps have been stacked so that each subcell or layer can absorb a different part of the wide energy distribution in the sunlight. This situation is advantageous, since each photon absorbed in a subcell corresponds to one unit of charge that is collected at the subcell operating voltage, which is approximately linearly dependent on the band gap of the semiconductor material of the subcell. An ideally efficient solar cell would have a large number of subcells, each absorbing only photons of energy negligibly greater than its band gap.
The most efficient and therefore dominant technology in multijunction photovoltaic cells is 2- and 3-junction GaInP/Ga(In)As/Ge cells. These monolithic cells are grown lattice matched to GaAs or Ge substrates. While variations on this material system, such as AlInGaP or lattice mismatched GaInP top cells, might provide a more ideal match of band gaps to the solar spectrum, practical considerations have indicated that lattice matched GaInP is preferred for large scale production. Addition of even small amounts of aluminum to the top cell to form AlInGaP simultaneously incorporates oxygen and thus quickly degrades the minority carrier lifetime and performance of the device. Lattice mismatched GaInP top cells induce dislocation formation having a similar effect.
In monolithic, series-interconnected, 2- and 3-junction GaInP/Ga(In)As/Ge solar cells, it is desirable for the GaInP top subcell to have nearly the same photogenerated current density as the Ga(In)As subcell. If the currents are different, the subcell with the lowest photogenerated current will limit the current through all of the series-interconnected subcells in the multijunction (MJ) cell, and excess photogenerated current in other subcells is wasted. Limiting the current in this manner results in a severe penalty on the MJ cell efficiency.
At the lattice constant of Ge (or of GaAs) substrates, GaInP grown under conventional conditions has an ordered group-III sublattice and therefore has a band gap which is too low to achieve the desired current match between subcells in the unconcentrated or concentrated AM0 space solar spectrum, the unconcentrated or concentrated AM1.5D and AM1.5G terrestrial solar spectra, and other solar spectra, unless the top subcell is purposely made optically thin, as in U.S. Pat. No. 5,223,043. To achieve the highest efficiencies, the thickness of the subcells in MJ cells are tuned to match the current in each subcell. As may be appreciated from the initial discussion of multijunction solar cells, it would be preferable to do the current matching by increasing the band gap of the top cell rather than reducing its thickness, producing a higher voltage at the same current. An important property of GaInP is that its band gap varies with growth conditions. GaInP grown under conventional conditions is GaInP with a CuPtB ordered group-III sublattice. The result of this ordering may be a decrease in band gap of up to 470 meV for completely ordered material compared with completely disordered material. A. Zunger, MRS Bulletin, 22, (1997) p. 20-26. Typically, this loss in band gap is only 120 meV since the ordering is only partial. The amount of ordering contained in a sample is described by the order parameter, η, which ranges from 0 (disordered) to 1 (completely ordered). G. B. Stringfellow, MRS Bulletin, 22, (1997) p. 27-32.
If the GaInP top cell is fully disordered, an optically thick top cell is nearly current matched for the AM1.5D and AM1.5G terrestrial spectra, but still must be slightly optically thin to match the AM0 spectrum. The increase ΔEg in band gap results in an increase in open-circuit voltage Voc of approximately ΔEg/q (typically 100 mV) for fully disordered GaInP as compared to partially ordered GaInP.
A conventional process 100 of forming a MJ cell is shown in FIG. 1. In a step 110, lower layers, including a buffer layer and a middle cell, in a structure are grown. Next, in a step 120, an n-type side of a tunnel junction (TJ) n++ layer is grown. Next, in a step 130, the p++ side of the TJ, a back surface field layer, and a GaInP top cell are grown. In a step 140, a window layer is grown. Next, in a step 150, the process is paused. Finally, in a step 160, a cap layer is grown. Under typical growth conditions, process 100 results in a MJ cell having the disadvantages identified herein.
The tuning of the band gap by controlling CuPt-type ordering has been studied. Early on, G. B. Stringfellow, “Order and Surface Processes in III-V Semiconductor Alloys,” MRS Bulletin, July 1997, p. 27-32 concluded that there was a one-to-one relationship between the degree of order and phosphorus (P) dimers on the surface at a growth temperature between 620-720° C. and partial pressure of <200 Pa. Additionally, it has long been known that growth at sufficiently high temperature would produce disordered GaInP. However, the high growth temperature required is not necessarily compatible with the growth of the complex multijunction cells on Ge substrates or may exceed those attainable by commercially available MOVPE reactors.
C. M. Fetzer et al., “The use of a surfactant (Sb) to induce triple period ordering in GaInP,” Appl. Phys. Lett., Vol, 76, No. 11, 13 Mar. 2000 indicated that Sb (and Bi) had been previously added as an isoelectronic surfactant during the growth of GaInP to alter the surface by replacing P dimers with Sb (or Bi) and to eliminate ordering. The term “isoelectronic” in relation to P was to describe the fact that Sb and P had the same number of valence electrons, the absence of first order changes in the Fermi level of the GaInP layer, and the lack of incorporation into the GaInP surface, which is in contrast to “dopant” surfactants.
G. B. Stringfellow et al., “Surface processes in OMVPE—the frontiers,” Journal of Crystal Growth 221 (2000) 1-11 used the surfactants Te (a donor) and As, Sb, and Bi (isoelectronic with P) for GaInP grown by organometallic vapor-phase epitaxy (OMVPE). Stringfellow et al. reported that each of the surfactants produced disordered layers under conditions that would normally produce highly ordered GaInP. It was suggested that As and Sb operated by surface changes, Te operated by kinetic effects, and Bi operated by both surface changes and kinetic effects.
More recently, C. M. Fetzer et al., “Effect of surfactant Sb on carrier lifetime in GaInP epilayers,” J. Appl. Phys., Vol. 91, No. 1, 1 Jan. 2002 grew Ga0.52In0.48P on GaAs using Sb as a surfactant at various concentrations. At an intermediate concentration (Sb/III(v)=0.016), CuPt ordering decreased and band gap energy increased. At a higher concentration (Sb/III(ν)=0.064), band gap energy decreased due to the onset of composition modulation.
Similarly, S. W. Sun et al., “Isoelectronic surfactant-induced surface step structure and correlation with ordering in GaInP,” Journal of Crystal Growth 235 (2002) 15-24 studied the surfactant concentration effects of As, Sb, and Bi. They noted that the intentional modulation of ordering could be used to fabricate heterostructures and quantum wells without changing the compositional material. In particular, the process could be used for solar cells, diodes, and lasers.
A disorder-on order-on disorder heterostructure using the surfactant Sb on GaInP was described by J. K. Shurtleff et al., “Time dependent surfactant effects on growth of GaInP heterostructures by organometallic vapor phase epitaxy,” Journal of Crystal Growth 234 (2002) 327-336. The heterostructures grown with interruptions in growth produced thin ordered layers and sharp interfaces.
In all of the above prior art describing the effect of isoelectronic surfactants on disordering GaInP, the experiments were carried out at an overall system pressure of 760 torr, or atmospheric pressure. Practical manufacture of high-volume and large-area epitaxy is optimally performed at low pressure to achieve the best uniformity and highest throughput. This difference in pressure is significant in that the hydrodynamics are vastly different between the two growth regimes. These differences are highlighted by Kikkawa et al., “Ordered-InGaPSb—GaAs-based FET and HBT structures grown by MOVPE,” 2001 Int. Conf. Proc on InP and Related Materials (2001) p. 464 who conclude that Sb surfactant disordering did not work at low pressures below ˜100 torr. Although this does not prove that the disordering does not function at low pressure, it does illustrate that the growth regimes used in research and manufacture are very different. Specifically, reduction of an idea functional in research to practice on manufacturing scales may require significant innovation.
As can be seen, there is a need for a process of making an optoelectronic device with a specific GaInP material system having a top subcell that converts photogenerated current densities at as high a voltage as possible and thereby maximizes the efficiency of the photovoltaic cell. Furthermore, there is a need for a process of making a photovoltaic cell with a specific GaInP material system having higher values of Voc and Vmp. Such a photovoltaic cell preferably includes a top subcell having the highest possible band gap achievable for lattice matched GaInP.